Fractional Lévy stable motion with LRD for RUL and reliability analysis of li-ion battery. (June 2022)

• Record Type: Journal Article
• Authors: Liu, He; Song, Wanqing; Zio, Enrico
• Is Part Of: ISA transactions. Volume 125(2022)
• Page Start: 360
• Page End: 370
• Publication Date: 2022-06
• Subjects: Fractional Lévy stable motion -- Remaining useful life -- Degradation model -- Long-range dependence -- Reliability; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2021.07.002
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: The Remaining Useful Life (RUL) is important for reliability analysis of li-ion battery. Reliability of li-ion battery decreases with shortened the RUL. The RUL of li-ion battery can be revealed by the capacity change. The future change of the capacity is related to the current and the historical states, namely, the capacity change of li-ion battery has Long-Range Dependence (LRD). This article describes a RUL prediction method based on fractional order Lévy stable motion (fLsm), which solves the LRD was not obvious caused by the excessive difference of the integer-order model. First, the LRD of the fLsm is revealed by stability index and integral kernel function with Hurst parameter. Then, the fLsm is used as a diffusion term, which reflects the stochastic and LRD of the RUL degradation, to establish a degradation prediction model. The iterative form of the prediction model is established through the incremental distribution of the fLsm. Finally, the RUL is predicted by the Monte Carlo simulation and degradation prediction model. The predictive performance of the fLsm degradation model is verified by battery data in different operating environments. The reliability of li-ion battery is analyzed by the RUL. Graphical abstract: Highlights: The capacity change of li-ion battery has LRD property. The fLsm can reveal LRD, heavy tail and non-Gaussian characteristics of li-ion battery degradation. The iterative difference form of the degradation model based on the fLsm is established. Compared with other RUL prediction model, the fLsm degradation model with power-law drift has better prediction performance.